High-Quality Short-Range Radar Imaging with Coprime Sampling

Short-range imaging radar, with its all-day and all-weather perception capabilities, has gained considerable attention in emerging fields such as autonomous vehicle sensing and industrial robotic perception. However, compared to traditional imaging radar, short-range imaging radar systems face more stringent constraints in terms of physical sampling resources, particularly the number of sampling channels and the resulting aperture size. These limitations lead to reduced resolution and a lower signal-to-noise ratio, ultimately degrading the imaging quality and making it difficult to interpret. To address these challenges, we explore coprime sampling as a strategy to achieve high-quality short-range radar imaging using limited physical sampling resources. Our approach is built upon three core perspectives: (1) physical sampling: we adopt a coprime pattern to form an extended sampling aperture with a structured layout, enabling effective utilization of limited channels and minimizing aperture loss; (2) signal measurement: we utilize the second-order statistics of the measured data to generate additional equivalent measurements, thus enhancing the system’s capability to capture diverse spatial information; and (3) scene reconstruction: we establish a novel forward measurement model, linking these equivalent measurements to the scene, and then formulate a sparsity-regularized optimization problem. We design a background-texture-preserving, target-enhanced resolving method based on the first-order proximal gradient algorithm to achieve robust and high-quality imaging results. Our method is verified on several measured data. The results show that our proposed approach achieves high-quality imaging while utilizing approximately half of the typical sampling resources. This study not only validates the effectiveness of coprime sampling for short-range radar imaging but also highlights its potential to alleviate sampling constraints in various resource-constrained applications.